New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Nightly

by yicheng47 in yicheng47/runner

Cut or check Runner nightlies — one workflow for macOS and Windows, one public prerelease, shared stamp, and separate stable feeds

MITAuto-check passedAgent Workflows

Install Nightly

skills CLI
$ npx skills add yicheng47/runner --skill nightly -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install yicheng47/runner nightly --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/yicheng47/runner.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/nightly .claude/skills/nightly && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
nightly
GitHub stars
275
Token cost
~2.2k tokens
SKILL.md length
1,155 words
Files
1
Skills in repo
4
Repo updated
First seen
Licence
MIT

At a glance

Cut or check Runner nightlies — one workflow for macOS and Windows, one public prerelease, shared stamp, and separate stable feeds

  • Works in 6 steps: Preflight — stop on any failure and… → Dispatch: gh workflow run nightly.yml… → Watch: gh run watch --exit-status. On… → …
  • Agent Workflows work in your project
  • SKILL.md covers Usage, run [platform], check [platform] and Notes
  • Calls gh, git and curl

What it does

Nightly is an agent skill from yicheng47/runner. Cut or check Runner nightlies — one workflow for macOS and Windows, one public prerelease, shared stamp, and separate stable feeds

Its SKILL.md is about 2.2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows. It works with macOS and Git. The repository describes itself as: Where terminal agents work together. Claude Code, Codex, Copilot CLI and pi on the same task, in one mission, each keeping its own TUI in a real terminal. The licence is MIT.

When your agent uses it

  • Agent Workflows work in your project

Example prompts

  • “/nightly”

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

  1. Preflight — stop on any failure and report it.
  2. Dispatch: gh workflow run nightly.yml --ref main for both, or append -f platform=macos / -f platform=windows. Find the new run with gh run…
  3. Watch: gh run watch --exit-status. On failure, inspect gh run view --log-failed and report the failing step. A failed/cancelled selected…
  4. Verify: run check for the selected platform(s), using this run's exact commit-based identity, stamp, and SHA. The shared identity is in…
  5. Record: nothing in docs/. The workflow run, the nightly tag and the release's assets and notes are the record of a cut…
  6. Report: shared identity, run result, download links, changes included, and any installed-upgrade checks still pending. macOS nightlies can…

What it can do on your machine

Read from SKILL.md and the folder at commit 4a4bbf0. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • gh
    • git
    • curl

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use gh, git and curl, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Nightly loads about 2.2k tokens when it runs. Until then it costs about 35 tokens; SKILL.md has 1,155 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~35
When it runs · the whole SKILL.md, loaded when a task matches
~2.2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from yicheng47/runner at commit 4a4bbf0, republished under its MIT licence (© yicheng47). 1,155 words, ~2,197 tokens.

Download SKILL.mdSave it as .claude/skills/nightly/SKILL.md (or your agent's skills folder).
name
nightly
description
Cut or check Runner nightlies — one workflow for macOS and Windows, one public prerelease, shared stamp, and separate stable feeds

Nightly

One nightly.yml dispatch builds both platforms by default from one SHA and UTC stamp. platform=macos or platform=windows selects one. One rolling public nightly prerelease serves macOS DMGs, a signed Sparkle appcast, Windows x64 installers, and matching minisign signatures. Each platform keeps two builds by stamp independently on that release. Production feeds and releases/latest stay separate. Contract: docs/arch/arch.md §14 and docs/arch/windows.md.

An explicit request to run authorizes that workflow dispatch. Reading or editing this skill and check do not authorize dispatch. Nightly is a rolling development channel identified by its commit; official version tags are separate, and no nightly crate-version bump is needed. Commit, push, merge, tagging, direct release mutation, app launch/restart, and installation require the user's explicit instruction. Record only observed results; never invent a successful cut or a PC smoke result.

Usage

/nightly [run [both|macos|windows] | check [both|macos|windows]] — no action means run; omitted platform means both.

run [platform]

  1. Preflight — stop on any failure and report it.
    • git fetch origin; require branch main, a clean tree (git status --short empty), and git rev-parse main origin/main equal. Capture the full origin/main SHA. Do not push local commits without instruction.
    • Check for any queued or running nightly, including other platforms: gh run list --workflow nightly.yml --limit 30 --json databaseId,status,displayTitle,headSha. Stop if a run is not completed; all platforms share concurrency group nightly, so a dispatch cancels the entire older run.
    • CI for the SHA: gh run list --commit <sha> --workflow ci.yaml --limit 1 --json status,conclusion. Missing or completed without success → stop. Queued/in-progress is fine: the single publish job waits for CI and verifies its successful conclusion. Both macOS and Windows CI must pass even for a single-platform cut.
    • If the Runner MCP is reachable, report running missions from mission_list_summary. A successful cut can produce a nightly update offer and a Windows background download. Installation/restart remains a separate user action and interrupts live PTYs.
  2. Dispatch: gh workflow run nightly.yml --ref main for both, or append -f platform=macos / -f platform=windows. Find the new run with gh run list --workflow nightly.yml --branch main --limit 5 --json databaseId,status,headSha,displayTitle,createdAt; require the selected Nightly (<platform>) title, creation after dispatch, and headSha equal to preflight. If the source changed during dispatch, report the mismatch and do not claim the intended cut succeeded.
  3. Watch: gh run watch <id> --exit-status. On failure, inspect gh run view <id> --log-failed and report the failing step. A failed/cancelled selected build prevents publication of both platforms on a both run. Failure or cancellation during publication can leave some files uploaded; report the cut incomplete and do not automatically retry.
  4. Verify: run check for the selected platform(s), using this run's exact commit-based identity, stamp, and SHA. The shared identity is in prepare's build-identity outputs/log and the successful publication job summary. Require the whole run to have succeeded, not just one build job.
  5. Record: nothing in docs/. The workflow run, the nightly tag and the release's assets and notes are the record of a cut; docs/impls/archive/504-nightly-channel-unification.md is the archived program record and keeps only its own first-cut section. Report installed upgrade/PC results only after observing them or receiving them from Jason.
  6. Report: shared identity, run result, download links, changes included, and any installed-upgrade checks still pending. macOS nightlies can update through Sparkle; Windows nightlies use the in-app updater. Do not install or restart either app from this skill without explicit instruction.
Show full SKILL.md (595 more words)Show less

check [platform]

This is read-only and does not dispatch. Read recent nightly.yml runs with gh run list --workflow nightly.yml --branch main --limit 10 --json databaseId,status,conclusion,headSha,createdAt,displayTitle. A both run applies to both platforms; a later single-platform cut may mean their assets now have different stamps on the same release. The nightly tag points at the last published cut, so git fetch origin refs/tags/nightly:refs/tags/nightly then git log nightly..origin/main --oneline describes what a new cut would pick up; the release notes list what the current cut has beyond the latest official release. A single-platform cut moves the tag too, so an unselected platform’s assets can be older than the tag: read their stamps from the filenames.

Read gh release view nightly --json isPrerelease,isDraft,assets once; require isPrerelease=true, isDraft=false, every selected platform’s expected filenames, and at most two timestamped installers and two DMGs independently. Require the shared notes from script/nightly-release-notes.md when checking the publish workflow; release notes cover installation and updates on both platforms. Sort by trailing YYYYMMDD.HHMM stamp, not by the version prefix. Both apps display Nightly (<sha>), independently of the crate version. The UTC stamp orders artifacts and updates; the short commit identifies their source. For example, a Windows artifact identity is nightly.abc1234.20260908.0100.

  • macOS / nightly: require Runner-Nightly-<sha>.<stamp>-arm64.dmg and appcast.xml. Check anonymous DMG access with curl --silent --show-error --fail --head --location <asset-url>; download and parse the appcast. Its single signed enclosure must point to that DMG at https://github.com/yicheng47/runner/releases/download/nightly/, with the expected stamp, seven-character commit as its short version, and arm64 hardware requirement. Sparkle supplies the app name in its native alert; Runner’s own update display is Nightly (<sha>). The release page is https://github.com/yicheng47/runner/releases/tag/nightly.
  • Windows / nightly: require Runner-Setup-nightly.<sha>.<stamp>-x64.exe and its exact .sig; anonymous access must succeed for both at https://github.com/yicheng47/runner/releases/download/nightly/. Installers/signatures prune together. Portable ZIPs are retired. The release page is https://github.com/yicheng47/runner/releases/tag/nightly.
  • Stable isolation: verify gh api repos/yicheng47/runner/releases/latest --jq .tag_name still identifies the current production release (0.8.2 when #502 was implemented), not a nightly. Do not report a real installed update or production Sparkle check as tested merely because release metadata is correct.

Notes

  • Nightlies are dispatch-only: a push to main runs CI alone. Nightlies need no crate-version bump or relationship to the next official version. One publish job validates selected artifacts, gates on CI once, uploads the DMG before its appcast and the installer before its signature, checks public downloads, then prunes. All uploads target nightly, and one public verification reads that release before either platform prunes. A single-platform cut leaves the other platform’s assets/feed and retention untouched. An upload or download failure reports the cut incomplete and prevents all pruning.
  • Nightly and production use com.wycstudios.runner / Runner.app on macOS and retain ~/Library/Application Support/com.wycstudios.runner/. Installing a nightly over Runner switches it to the nightly feed; installing a stable DMG switches it back. Windows follows the same install-to-switch rule. Public prereleases can be found by browsing, while production installs read their separate stable feed.
  • One-time transition: Jason installs the first unified Mac DMG over Runner.app, then deletes Runner Nightly.app by hand; its old bundle id cannot update through Sparkle to the new one. The PC also needs a manual install because its existing build reads nightly-win. After the PC reads nightly, Jason deletes the old nightly-win release and tag by hand. There is no transition mirror; neither run nor check reads or writes that release. Dormant Windows installs reinstall from nightly after it is deleted. These manual steps still require explicit authorization.
  • Production releases remain a different path: tag vX.Y.Z at a bare X.Y.Z crate version → release.yml builds a draft; publishing is the human's switch. Do not tag from this skill.

© yicheng47, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .agents/skills/nightly of yicheng47/runner.

Open the folder on GitHubat commit 4a4bbf0

Compare with similar skills

Nightly next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Nightly compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Nightly this skillyicheng47/runner275—~2.2kAutomated safety check: PassMIT
Codexless Release Supervisorliyana31811/Codexless118—~2.4kAutomated safety check: PassApache-2.0
Build And Run macOS Approbinebers/openusage4.3k—~473Automated safety check: PassMIT
Build Px4 macOSPX4/PX4-Autopilot13k—~1.1kAutomated safety check: PassBSD-3-Clause
Git Worktree Setupk8zdev/k8z155—~930Automated safety check: PassNone
Mori Agent Bridgevaayne/mori303—~1.7kAutomated safety check: PassMIT

Similar skills

  • Codexless Release Supervisor

    liyana31811/Codexless

    Prepare, validate, and publish Codexless preview/hotfix releases from the canonical household source, including acceptance anti-omission gates, candidate provenance/parity, manifest/build identity…

    118 GitHub stars~2.4k tokensUpdated 11 days ago
    Agent WorkflowsAuto-check passed

  • Build And Run macOS App

    robinebers/openusage

    Slash command that creates or updates the project-local macOS script/buildandrun.sh and uses it as the default build/run entrypoint.

    4.3k GitHub stars~473 tokensUpdated today
    Agent WorkflowsAuto-check passed

  • Build Px4 macOS

    PX4/PX4-Autopilot

    Build PX4 board firmware on macOS in the px4-dev Docker container, including git worktrees, and stage commit-labeled artifacts without flashing hardware.

    13k GitHub stars~1.1k tokensUpdated yesterday
    DevelopmentAuto-check passed

  • Git Worktree Setup

    k8zdev/k8z

    Automate git worktree creation and copy private files (not tracked in git) to new workdirs.

    155 GitHub stars~930 tokensUpdated 3 mo ago
    DevelopmentAuto-check passed

  • Mori Agent Bridge

    vaayne/mori

    Use Mori CLI to enable agent-to-agent communication and coordination across Mori-managed panes.

    303 GitHub stars~1.7k tokensUpdated 2 mo ago
    DevelopmentAuto-check passed

  • Swiftui Patterns

    robinebers/openusage

    Best practices and example-driven guidance for building native macOS SwiftUI scenes and components, including windows, commands, toolbars, settings, split views, inspectors, menu bar extras, and…

    4.3k GitHub stars~3.4k tokensUpdated today
    MobileAuto-check passed

More from yicheng47/runner

  • Bug

    yicheng47/runner

    Report, list, or manage bug reports as GitHub issues. An agent skill from yicheng47/runner.

    275 GitHub stars~949 tokensUpdated today
    Auto-check passed

  • Release

    yicheng47/runner

    Cut a Runner production release — bump the workspace version, tag vX.Y.Z, let release.yml build the draft for both platforms, write bilingual release notes (English, with 中文 collapsed in a details…

    275 GitHub stars~2.4k tokensUpdated today
    Auto-check passed

  • Feature

    yicheng47/runner

    File, list, or manage feature issues, and write a feature's spec when its development starts

    275 GitHub stars~1.3k tokensUpdated today
    Auto-check passed

Works with

Categories

Questions about Nightly

What does Nightly do?

Cut or check Runner nightlies — one workflow for macOS and Windows, one public prerelease, shared stamp, and separate stable feeds. Nightly is an agent skill from yicheng47/runner.

When should I use Nightly?

Nightly fits situations like: agent Workflows work in your project.

How do I install Nightly in Claude Code?

Run `npx skills add yicheng47/runner --skill nightly -a claude-code`. Or copy the skill folder (.agents/skills/nightly in yicheng47/runner) into .claude/skills/nightly in your project. Claude Code loads it when a task matches its description.

How do I install Nightly in Codex?

Run `npx skills add yicheng47/runner --skill nightly -a codex`. Or copy the skill folder (.agents/skills/nightly in yicheng47/runner) into .agents/skills/nightly in your project. Codex loads it when a task matches its description.

Can I use Nightly in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add yicheng47/runner --skill nightly -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/nightly, .gemini/skills/nightly, .github/skills/nightly and .opencode/skills/nightly in your project.

What does Nightly need to run?

Going by SKILL.md and its folder, Nightly needs the command-line tools its instructions call (gh, git and curl).

Does Nightly access the network?

SKILL.md contains no URLs. Its commands use gh, git and curl, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Nightly safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Nightly use?

Nightly is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Nightly use?

About 2.2k tokens (SKILL.md is roughly 8.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Nightly?

Skills that share tags, products or a category with Nightly: Codexless Release Supervisor (liyana31811/Codexless, 118 stars), Build And Run macOS App (robinebers/openusage, 4.3k stars), Build Px4 macOS (PX4/PX4-Autopilot, 13k stars) and Git Worktree Setup (k8zdev/k8z, 155 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Nightly?

yicheng47 (a GitHub user) maintains it in yicheng47/runner, which has 275 GitHub stars. The repository holds 4 skills in this directory. The repository was last updated on October 10, 2026.

Source: yicheng47/runner on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.